MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202631087124 A) filed by National Institute Of Technology Jamshedpur on July 16, 2026, for System And Method For Early-Stage Alzheimer'S Disease Detection Using Hybrid Multimodal Deep Learning Framework.
Inventors include Kumar, Rahul; Gavel, Harish; Azad, Chandrashekhar; and Shaw, Dilip Kumar.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: SYSTEM AND METHOD FOR EARLY-STAGE ALZHEIMER'S DISEASE DETECTION USING HYBRID MULTIMODAL DEEP LEARNING FRAMEWORK Embodiment of the present disclosure relates to a system (100) and method (300) for automated early-stage detection and progression prediction of Alzheimer's Disease. The method comprises processing (302) multimodal patient data to obtain normalized multimodal patient data. Further, the method comprises extracting (304A) modality-specific spatial feature vectors and determining (304B) a temporal disease progression pattern from the spatial feature vectors obtained from a plurality of patient visits. Thereby, the method comprises combining (306) the modality-specific spatial feature vectors and the temporal disease progression pattern using an attention-based fusion mechanism to obtain fused multimodal feature data. Moreover, the method comprises dynamically adjusting (308) modality contribution weights and classifying (310) the fused multimodal feature data into Alzheimer's Disease progression categories. Further, the method comprises determining (312) an interpretable prediction output and providing (314) a diagnostic output comprising classification results, confidence information, conversion risk probability, and uncertainty information.
Disclaimer: Curated by HT Syndication.
